Method and apparatus for reprogramming protein functional domain based on tee server cluster

Abstract

The application relates to a protein function domain reprogramming method and device based on a TEE server cluster. The method comprises the following steps: receiving protein data to be reprogrammed containing a known function domain sent by a client; performing similar domain retrieval on the known function domain based on a TED database to obtain a target function domain similar in structure; obtaining target replacement protein data corresponding to the target function domain from a protein database; performing protein data correlation comparison and function domain boundary adjustment, completing function domain replacement processing according to an adjustment result, generating reprogrammed protein data and returning the reprogrammed protein data to the client. By adopting the TEE hardware-level security feature and a multi-algorithm collaborative mechanism, the method realizes data security and operation credibility of the whole process of protein function domain reprogramming, and effectively improves the accuracy and reliability of protein function domain replacement.

Description

Technical Field

[0001] This application relates to the field of protein engineering technology, and in particular to a method and apparatus for protein functional domain reprogramming based on a TEE server cluster. Background Technology

[0002] With the deepening of research in protein engineering and structural biology, protein domain reprogramming technology has become an important means of modifying protein function and designing novel biological agents. By replacing key functional domains of proteins, new activities can be selectively endowed, stability improved, or binding specificity optimized. This provides new ideas for solving problems such as the treatment of drug-resistant diseases and the low catalytic efficiency of industrial enzymes. Its technological breakthroughs are directly related to the innovation capabilities of the biopharmaceutical industry and the progress of life science research.

[0003] In traditional technological paradigms, protein domain reprogramming typically relies on structural similarity searches of known protein functional domains and alignment with target proteins to achieve precise domain replacement. These processes are usually executed on ordinary servers or cloud computing platforms. However, current methods present technical challenges regarding data security and computational reliability. Specifically, firstly, there is the data security bottleneck: the Encyclopedia of Domains (TED) database and client-side private protein data (such as target protein data to be reprogrammed) upon which domain reprogramming depends are vulnerable to data theft in traditional computing environments. In particular, the leakage of core data such as protein functional domain ranges and structural similarity search results could lead to infringement of research findings or misuse of biometric information. Secondly, there is the computational reliability bottleneck: the domain reprogramming process uses critical operations such as structural similarity searches. In untrusted environments, algorithm parameters may be tampered with or computation results may be hijacked, causing the reprogrammed protein function to deviate from expectations. This significantly increases research costs and application risks, forcing research institutions to struggle to balance technical efficiency and data security. Summary of the Invention

[0004] Therefore, it is necessary to provide a method and apparatus for protein functional domain reprogramming based on a server cluster using a Trusted Execution Environment (TEE) to address the aforementioned technical problems and ensure data security and computational reliability during the protein functional domain reprogramming process.

[0005] In a first aspect, this application provides a protein functional domain re-editing method based on a TEE server cluster, wherein the TEE server cluster stores a TED database, a protein sequence database, and a protein structure database, and the method is executed by the TEE server cluster, including:

[0006] Receive protein data to be reprogrammed sent by the client, wherein the protein data to be reprogrammed contains information about known functional domains;

[0007] Based on the TED database, a similarity search is performed on the known functional domains of the protein data to be recompiled to obtain target functional domains that are similar to the known functional domains.

[0008] Obtain target substitution protein data corresponding to the target functional domain from the protein sequence database and the protein structure database;

[0009] The protein data to be reprogrammed is correlated and compared with the target replacement protein data, and the boundary of the target functional domain of the target replacement protein data is adjusted to obtain the target functional domain boundary adjustment result.

[0010] Based on the target functional domain boundary adjustment results, the known functional domains in the protein data to be reprogrammed are replaced to generate functional domain reprogrammed protein data.

[0011] The protein data after the functional domains have been reprogrammed is returned to the client.

[0012] In one embodiment, based on the TED database, a similarity search is performed on known functional domains of the protein data to be recompiled to obtain target functional domains similar to the known functional domains, including:

[0013] The master node of the TEE server cluster splits the similarity retrieval task into a preliminary screening sub-task based on the FoldSeek algorithm and a fine screening sub-task based on the TM-align algorithm, and distributes them to different slave nodes.

[0014] The slave nodes that perform the initial screening subtask use the FoldSeek algorithm to perform the initial comparison and obtain the initial screening results of similar functional domains;

[0015] The slave node executing the fine screening task then uses the TM-align algorithm to perform a second comparison of the initial screening results, and selects the target functional domains whose similarity to the known functional domains reaches a preset threshold.

[0016] The master node summarizes the retrieval results of the slave nodes that performed the fine screening task.

[0017] In one embodiment, the master node splits the similar functional domain retrieval task into a preliminary screening sub-task based on the FoldSeek algorithm and a fine screening sub-task based on the TM-align algorithm, and distributes them to different slave nodes, including:

[0018] The master node generates a first temporary session key for each of the slave nodes and transmits the first temporary session key to the slave nodes through asymmetric encryption.

[0019] A dual-layer encryption mechanism is adopted, which uses the first temporary session key to encrypt the task information of the primary screening task and the secondary screening task, and performs AES encryption on the known protein functional domains;

[0020] The encrypted task information and the known protein functional domains are distributed to each slave node.

[0021] In one embodiment, the protein data to be reprogrammed is correlated and compared with the target replacement protein data, and the target functional domain boundary of the target replacement protein data is adjusted to obtain the target functional domain boundary adjustment result, including:

[0022] The master node of the TEE server cluster splits the association comparison task into multiple comparison sub-tasks and distributes them to each slave node.

[0023] Each slave node calls the Dali distance matrix comparison algorithm to construct a distance matrix and align the structure of the target replacement protein and the protein to be reprogrammed.

[0024] Based on the results of the Dali distance matrix alignment algorithm, the spatial position matching degree of functional domain boundary residues is analyzed, the functional domain boundary intervals of the target replacement protein are adjusted, and a functional domain boundary adjustment result set is obtained.

[0025] In one embodiment, after adjusting the functional domain boundary regions of the target replacement protein to obtain a functional domain boundary adjustment result set, the method further includes:

[0026] The master node summarizes the functional domain boundary adjustment result set of each slave node;

[0027] The master node adopts a multi-node cross-validation strategy to perform consistency verification on the functional domain boundary adjustment results obtained by distributing the same comparison task to different slave nodes, and obtains the functional domain boundary adjustment results whose consistency reaches a preset threshold.

[0028] In one embodiment, the master node of the TEE server cluster splits the comparison task into multiple comparison sub-tasks and distributes them to each slave node, including:

[0029] The master node generates a second temporary session key for each of the slave nodes and transmits the second temporary session key to the slave nodes through the cluster internal key exchange mechanism;

[0030] A layered encryption mechanism is adopted, in which the task information of the alignment subtask is encrypted with a public key, and the target replacement protein and the protein to be reprogrammed are encrypted with the second temporary session key.

[0031] The encrypted task information, the target replacement protein, and the protein to be reprogrammed are distributed to each slave node.

[0032] In one embodiment, the TED database, the protein sequence database, and the protein structure database are stored in the TEE server cluster using a distributed encrypted storage strategy, the distributed encrypted storage strategy including:

[0033] The master node of the TEE server cluster uses the Gocryptfs tool to encrypt the TED database, the protein sequence database, and the protein structure database, and performs fragmentation processing to obtain encrypted fragmented data.

[0034] The master node stores the encrypted shard data to the slave node and saves an encrypted storage mapping table in the master node's TEE environment. The storage mapping table records the correspondence between the encrypted shard data and the slave node.

[0035] Secondly, this application also provides a protein functional domain re-editing device based on a TEE server cluster, wherein the TEE server cluster stores a TED database, a protein sequence database, and a protein structure database, and the device is executed by the TEE server cluster. The device includes:

[0036] The data receiving module is used to receive protein data to be reprogrammed sent by the client, wherein the protein data to be reprogrammed contains information about known functional domains;

[0037] The similarity domain retrieval module is used to perform similarity retrieval on the known functional domains of the protein data to be recompiled based on the TED database, and to obtain target functional domains that are similar to the known functional domains.

[0038] The protein acquisition module is used to acquire target substitution protein data corresponding to the target functional domain from the protein sequence database and the protein structure database.

[0039] The boundary adjustment module is used to correlate and compare the protein data to be recompiled with the target replacement protein data, adjust the target functional domain boundary of the target replacement protein data, and obtain the target functional domain boundary adjustment result.

[0040] The domain replacement processing module is used to replace the known functional domains in the protein data to be recompiled according to the target functional domain boundary adjustment results, and generate functional domain recompiled protein data.

[0041] The result return module is used to return the protein data after the functional domain reprogramming to the client.

[0042] In one embodiment, the apparatus further includes the client and a third-party verification server:

[0043] The client is used to encrypt the protein data to be recompiled, initiate a functional domain recompilation operation request to the TEE server cluster, and obtain the protein data after functional domain recompilation.

[0044] The third-party verification server is used to receive and verify the overall trustworthiness certificate of the TEE server cluster sent by the TEE server cluster to confirm the trustworthiness of the TEE server cluster.

[0045] The aforementioned protein functional domain reprogramming method and apparatus based on a TEE server cluster deploys the protein reprogramming computation process within the TEE cluster. After receiving encrypted data, it employs a multi-algorithm collaborative retrieval mechanism to improve the accuracy and completeness of similar functional domain screening, utilizes distance comparison to adjust functional domain boundaries, and ultimately completes secure and reliable functional domain replacement. This method, through the hardware-level security features of TEE and the clustered collaborative mechanism, ensures data security and computational reliability throughout the entire functional domain reprogramming process, while guaranteeing that protein data is not leaked and critical operations are not tampered with. This effectively solves the problem of balancing security and reliability in traditional technologies. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a diagram illustrating the application environment of a protein functional domain reprogramming method based on a TEE server cluster in one embodiment.

[0048] Figure 2 This is a flowchart illustrating a protein functional domain reprogramming method based on a TEE server cluster in one embodiment.

[0049] Figure 3 This is a flowchart illustrating a known functional domain similarity retrieval method in one embodiment;

[0050] Figure 4 This is a flowchart illustrating a similar functional domain retrieval task allocation method in one embodiment;

[0051] Figure 5 This is a flowchart illustrating a method for comparing and adjusting the boundary of a target functional domain in one embodiment.

[0052] Figure 6 This is a schematic diagram of the functional domain boundary adjustment result based on the Dali algorithm in one embodiment;

[0053] Figure 7 This is a flowchart illustrating a task allocation method in one embodiment;

[0054] Figure 8 This is a structural block diagram of a protein functional domain reprogramming device based on a TEE server cluster in one embodiment.

[0055] Figure 9 This is a schematic diagram of the process of obtaining results through an authorization mechanism in one embodiment. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0057] The protein functional domain reprogramming method based on a TEE server cluster provided in this application can be applied to, for example... Figure 1 In the application environment shown, client 102 communicates with TEE server cluster 104 via a network. TEE server cluster 104 internally stores the TED database, protein sequence data, and protein structure data necessary for performing functional domain recompilation. Client 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, personal workstations, laboratory-specific clients, or other devices requiring complex biological data computation and analysis. Users submit protein data to be recompiled, containing known functional domain information, through client 102 and ultimately receive the recompiled results. The TEE server cluster 104 is a distributed system composed of multiple server nodes equipped with TEE hardware technology. This cluster can be a physical server cluster deployed in a private or public cloud environment, or a cloud server cluster specifically built for biological computing services, providing secure TEE computing capabilities.

[0058] In one exemplary embodiment, such as Figure 2 As shown, a protein functional domain reprogramming method based on a TEE server cluster is provided, which is then applied to... Figure 1 Taking the TEE server cluster as an example, the explanation includes the following steps S201 to S206. Wherein:

[0059] Step S201: Receive the protein data to be reprogrammed sent by the client. The protein data to be reprogrammed contains information about known functional domains.

[0060] It should be noted that protein domain reprogramming refers to the technique of altering the functional properties of a protein by replacing specific functional regions. In this embodiment, the protein domain reprogramming process mainly includes four core steps: similar functional domain retrieval, target protein alignment, functional domain boundary adjustment, and functional domain replacement. This process relies on the support of specialized databases. These include the TED database, which provides functional domain classification information based on three-dimensional structural topology for rapid identification of structurally similar functional domains, as well as protein structure databases and protein sequence databases.

[0061] To ensure the security of critical data involved in protein functional domain reprogramming during storage and processing, and to prevent data leakage and tampering risks due to system vulnerabilities or malicious attacks, this invention is based on Trusted Execution Environment (TEE) technology. A TEE is a secure computing environment built using hardware isolation technology, ensuring that code and data executed within it will not be leaked or tampered with even if the system is attacked. In this embodiment, the aforementioned databases are all stored in encrypted form within a TEE server cluster. Specifically, encryption algorithms are used to encrypt the TED database, protein sequence data, and structural data, and a distributed storage strategy is employed to achieve distributed management and rapid data location.

[0062] In one specific implementation, the Gocryptfs tool is used to encrypt the database file. This tool is based on the AES-256-GCM algorithm and encrypts the data through file-level encryption. It also supports filename encryption, thereby ensuring the confidentiality of protein plaintext data by converting it into encrypted data that cannot be directly identified.

[0063] The encrypted data undergoes distributed storage operations through the master node of the TEE cluster. The master node distributes the encrypted data to each slave node according to preset rules (including data shard size, node load balancing strategy, etc.). At the same time, an encrypted storage mapping table is created in the master node's TEE environment. This mapping table records the correspondence between sharded data and slave nodes to achieve distributed data management and rapid data location. The data is stored long-term in the TEE cluster.

[0064] The protein data to be recompiled includes the full sequence and structural information of the proteins whose functional domains need to be replaced, as well as information on the known functional domains that need to be replaced. The information on the known functional domains includes key parameters such as the sequence range of the functional domain, its three-dimensional structural coordinates, and its functional annotations.

[0065] It should be noted that before transmitting data, the client needs to encrypt the re-edited protein data and known functional domains. Specifically, the data is encrypted using an encryption algorithm agreed upon with the TEE server to ensure it is encrypted before transmission. The client then transmits the encrypted protein data to be re-edited to the TEE server's master node through a secure transmission channel. This secure transmission channel is constructed using an encrypted transmission protocol to prevent unauthorized interception or tampering of data during transmission, ensuring data transmission security. Correspondingly, upon receiving the data, the TEE server needs to decrypt it within the TEE's secure environment.

[0066] Step S202: Based on the TED database, perform a similarity search on the known functional domains of the protein data to be recompiled to obtain target functional domains that are similar to the known functional domains.

[0067] In this step, the TEE server cluster can employ a multi-algorithm collaborative retrieval mechanism to improve the accuracy and completeness of similar functional domain retrieval. In practical applications, protein structure similarity retrieval usually needs to balance efficiency and accuracy; therefore, a staged retrieval strategy can be adopted.

[0068] In one embodiment, the master node of the TEE server cluster coordinates multiple slave nodes within the TEE security environment to divide the retrieval task into two stages: In the first stage, some slave nodes execute a fast initial screening algorithm, which quickly compares the structural features of known functional domains with the structural data in the TED database to select a preliminary candidate set of potentially similar functional domains; In the second stage, another group of slave nodes executes a fine comparison algorithm, which performs a precise three-dimensional structural comparison of each functional domain in the preliminary candidate set, calculates the structural similarity index, and selects multiple similar functional domains that meet a preset threshold.

[0069] Optionally, the fast initial screening algorithm can employ a structural feature-based retrieval method, such as a fast alignment algorithm based on a structural alphabet; the fine alignment algorithm can employ a topological structure-based alignment algorithm, such as an alignment method based on a distance matrix. This dual-algorithm verification mechanism effectively solves the problem of missed or false detections that may occur with traditional single-algorithm retrieval.

[0070] It should be noted that the search results include a list of similar functional domains, as well as key information such as identification information (e.g., UniProt ID), functional domain intervals (start and end residue positions), and structural similarity scores. The functional domain interval refers to the start and end positions of the functional domain within its complete protein sequence, usually represented by residue numbers.

[0071] After each slave node completes its retrieval, it locally encrypts the result data and generates a hash value for integrity verification. These encrypted results and verification data are then returned to the master node through a trusted internal channel within the cluster, ensuring that the retrieval results are not tampered with or leaked during transmission.

[0072] After receiving the encrypted results from all slave nodes in the TEE environment, the master node decrypts, summarizes, and integrates them, removing duplicate operation entries to form a verified candidate list of similar functional domains. This candidate list not only contains information on structurally similar functional domains but also ensures data integrity and consistency, providing a reliable input basis for subsequent functional domain replacement operations. This dual-algorithm verification mechanism improves the accuracy and completeness of similar functional domain retrieval, solving the problem of missed detections that may occur with traditional single-algorithm retrieval.

[0073] Step S203: Obtain the target replacement protein data corresponding to the candidate list of similar functional domains from the protein sequence data and protein structure data.

[0074] It is understandable that the candidate list of similar functional domains contains the core information of verified similar functional domains. However, to re-edit functional domains, it is necessary to obtain the complete protein data to which these functional domains belong. Based on this, the TEE server cluster retrieves the corresponding complete target protein data from distributed protein sequence databases and protein structure databases based on the UniProtID identifier of each functional domain in the candidate list.

[0075] Specifically, the master node first parses the UniProtID of each functional domain in the candidate list, and then locates and queries the slave nodes storing the corresponding complete protein data based on a pre-established encrypted data mapping table. It should be noted that the UniProtID, as a unique identifier for a protein, can accurately pinpoint the location of a specific protein in the distributed storage system.

[0076] In one possible implementation, the retrieval process can employ a parallel query mechanism, where the master node distributes query tasks for multiple UniProt IDs to different slave nodes. Each slave node decrypts locally stored database fragments within a TEE secure environment, extracting the complete sequence information, three-dimensional structural coordinates, and annotation functions of the target protein.

[0077] Optionally, the query results may include the protein's full-length amino acid sequence, atomic coordinate matrix, secondary structure assignment, and functional classification information. This data is encrypted locally on the slave node before transmission and returned to the master node via a trusted internal channel within the cluster, ensuring the confidentiality and integrity of the intermediate data.

[0078] Step S204: The protein data to be recompiled is correlated and compared with the target replacement protein data, and the boundary of the target functional domain of the target replacement protein data is adjusted to obtain the boundary adjustment result of the target functional domain.

[0079] First, the master node calls the target replacement protein data and associates it with the protein data to be recompiled to form a comparison dataset, thereby establishing the correspondence between known functional domains and candidate functional domains.

[0080] Next, the master node breaks down the overall alignment task into several subtasks based on the number and structural complexity of the target replacement proteins. In a specific implementation, each subtask contains a set of alignment requirements between the target replacement proteins and the proteins to be reprogrammed, along with the necessary alignment parameter templates.

[0081] The algorithm executes a structure alignment algorithm within a TEE (Trusted Environment for Energy) secure environment, using distance matrix construction and dynamic programming to find the optimal structure alignment path. Optionally, a protein structure alignment method based on the inter-residue distance matrix, such as the Dali algorithm, can be employed. The Dali algorithm is a protein structure alignment method based on the inter-residue distance matrix that can accurately identify structurally similar regions. Specifically, the algorithm constructs a distance matrix between the functional domains of the known protein and the target substitution protein, and then uses dynamic programming to find the optimal structure alignment path. The output of this step includes the residue correspondence for structure alignment and a structural similarity score.

[0082] Next, based on the output of the structure alignment algorithm, functional domain boundaries are adjusted. In one embodiment, the structural continuity of the functional domain boundary residues is analyzed according to the residue correspondence and spatial matching information, and the boundary positions are expanded or contracted as necessary. For example, the adjustment range is usually controlled within 1-3 residues to ensure that the replaced functional domain can be smoothly connected to the original protein in structure, while maintaining the overall folding stability of the protein.

[0083] Finally, the boundary-adjusted results are aggregated from the slave nodes to the master node. It should be noted that all computational tasks in the above steps are performed within a TEE secure environment, ensuring the immutability of the comparison parameters and adjustment logic, and guaranteeing the security of intermediate results and final adjusted data.

[0084] Optionally, the boundary adjustment process can also be combined with a multi-node cross-validation mechanism. When the same comparison task is assigned to multiple slave nodes, the consistency of the output results of each node is required to reach a preset threshold, thereby further improving the reliability of the boundary adjustment results.

[0085] Step S205: Based on the target functional domain boundary adjustment results, replace the known functional domains in the protein data to be recompiled to generate functional domain recompiled protein data.

[0086] After obtaining the verified functional domain boundary adjustment results, the master node of the TEE server cluster performs the functional domain replacement operation within the TEE secure environment. Specifically, when the slave nodes encrypt and transmit the boundary adjustment results, the master node first decrypts the encrypted boundary adjustment results to obtain the precise functional domain boundary information of each target replacement protein. It should be noted that the decryption operation is performed only in the trusted memory of the TEE, ensuring that sensitive data is not exposed in plaintext to an insecure environment.

[0087] Next, the adjusted functional domains are replaced with the corresponding positions in the original protein, generating functional domain-recompiled protein data containing complete sequence and structural information. This data is encrypted within the TEE environment using the same encryption algorithm as used during the initial data transmission on the client side, ensuring data security in subsequent transmission stages.

[0088] Step S206: Return the functional domain-recompiled protein data to the client.

[0089] After verifying the client's identity through a secure authorization mechanism, the TEE server cluster transmits the recompiled protein data to the client via a secure channel.

[0090] In some specific implementations, a multi-layered security mechanism is employed to achieve authorized data transmission. The client first initiates an authorization request to the TEE server cluster master node, submitting its identity identifier and task credentials. The master node verifies the client's identity through the TEE remote authentication mechanism, and after successful verification by a third-party verification server, generates a one-time encrypted authorization credential and returns it to the client. It should be noted that this authorization credential is time-sensitive and unique, and is bound to a specific computational task, effectively preventing credential misuse or replay attacks. After obtaining the authorization credential, the client can initiate a result retrieval request to the master node. After verifying the validity and scope of permissions of the credential, the master node transmits the encrypted functional domain recompilation result to the client through a secure transmission channel.

[0091] Finally, the client decrypts the received encrypted result using a preset decryption key in its local environment, ultimately obtaining usable functional domain recompiled protein data.

[0092] The protein functional domain reprogramming method based on TEE servers provided in the above embodiments achieves full lifecycle security protection for data processing by deploying the protein reprogramming computational processing flow in a TEE server cluster. Specifically, the confidentiality of data transmission, storage, and processing is ensured through the TEE hardware-level security environment; the accuracy of functional domain retrieval and boundary optimization is improved through a multi-algorithm collaborative mechanism, solving the technical problem of balancing security and accuracy in traditional protein reprogramming methods.

[0093] In one exemplary embodiment, such as Figure 3 As shown, in step S202 above, based on the TED database, a similarity search is performed on the known functional domains of the protein data to be recompiled to obtain target functional domains similar to the known functional domains. This specifically includes the following steps S301 to S304:

[0094] In step S301, the master node of the TEE server cluster splits the similarity retrieval task into a preliminary screening task based on the FoldSeek algorithm and a fine screening task based on the TM-align algorithm, and distributes them to different slave nodes.

[0095] It should be noted that the task splitting process is based on the structural domain classification index of the TED database (such as all-alpha structure, all-beta structure, alpha+beta structure, etc.) and the real-time computing power status of the slave nodes for dynamic planning, splitting the similar functional domain retrieval task into FoldSeek initial screening subtasks and TM-align fine screening subtasks. In one embodiment, the master node divides the retrieval task into multiple subsets according to the structural domain type, where the FoldSeek subtask is assigned to the high-concurrency processing node group, and the TM-align subtask is assigned to the high-precision computing node group. Simultaneously, additional task execution parameters are added, such as the e-value threshold of the FoldSeek algorithm and the minimum alignment length of the TM-align algorithm. Optionally, a double-layer encryption mechanism is used for task distribution to ensure the confidentiality of subtask data during transmission.

[0096] Step S302: The slave nodes performing the initial screening sub-task use the FoldSeek algorithm to perform the initial comparison and obtain the initial screening results of similar functional domains.

[0097] Understandably, the rapid initial screening stage requires quickly locating potentially similar functional domains within a large-scale database. In one specific implementation, the FoldSeek algorithm module is loaded from the slave node within the TEE environment, converting the three-dimensional structure of the known functional domains into a sequence of structural letters, which is then quickly compared with the encrypted structural data in the TED database. Optionally, the number of functional domains output by the algorithm can be adjusted by setting an e-value threshold.

[0098] In step S303, the slave node executing the fine screening task uses the TM-align algorithm to perform a second comparison of the initial screening results, and selects the target functional domains whose similarity to the known functional domains reaches a preset threshold.

[0099] The fine screening stage requires further structural verification of the initial screening results. Specifically, the TM-align algorithm is executed on each node within the TEE environment to calculate the TM-score and RMSD values ​​of each candidate functional domain compared to known functional domains. High-quality similar functional domains that meet preset TM-score criteria and structural alignment length requirements are selected. Optionally, the fine screening process can also incorporate functional domain annotation information for auxiliary filtering, ensuring that the final target functional domains possess both structural similarity and functional relevance.

[0100] Step S304: The master node summarizes the retrieval results of the slave nodes that performed the fine screening task.

[0101] Understandably, integrating distributed search results requires ensuring data integrity and consistency. The aggregation process is completed within a secure TEE environment, where the master node decrypts, deduplicates, and verifies the consistency of the encrypted results returned by each slave node.

[0102] In some specific embodiments, a dual-algorithm cross-validation mechanism can also be employed to ensure that each entry in the final candidate list of similar functional domains passes both the initial screening and the fine screening. For example, firstly, the master node parses the functional domain identifiers and their corresponding e-value scores from the initial screening results, and simultaneously obtains the TM-score and RMSD values ​​of the same functional domain identifiers from the fine screening results; secondly, the system applies dual-validation conditions to each candidate functional domain, requiring that the threshold criteria of e-value ≤ 0.01 and TM-score ≥ 0.5 be met simultaneously; finally, the system retains the functional domains that pass both algorithms. Optionally, for inconsistencies that occur during the validation process, the system can also record detailed algorithm parameters and comparison results to form a validation log for subsequent algorithm optimization.

[0103] Through the aforementioned phased, multi-algorithm collaborative retrieval mechanism, this embodiment effectively balances retrieval efficiency and accuracy while ensuring data security, providing a reliable set of functional domain candidates for subsequent functional domain replacement. Simultaneously, the distributed task execution method fully utilizes the computing resources of the TEE server cluster, significantly improving the retrieval performance of large-scale protein databases.

[0104] In practical applications, the task distribution process involves sensitive functional domain structure data, requiring strict security measures to prevent data leakage. Therefore, in some exemplary embodiments, such as... Figure 4 As shown, in step S301 above, the master node splits the similar functional domain retrieval task into a preliminary screening sub-task based on the FoldSeek algorithm and a fine screening sub-task based on the TM-align algorithm, and distributes them to different slave nodes, including the following encrypted transmission steps S401 to S403:

[0105] Step S401: The master node generates a first temporary session key for each slave node and transmits the first temporary session key to the slave node through asymmetric encryption.

[0106] It should be noted that the first temporary session key is time-sensitive and unique to prevent reuse or cracking. Specifically, the master node encrypts the first temporary session key using its public key, ensuring that only the target slave node can decrypt and obtain the key. Optionally, the key's validity period is set to a single task execution cycle, automatically expiring upon task completion.

[0107] Step S402: A two-layer encryption mechanism is adopted, which uses the first temporary session key to encrypt the task information of the primary screening task and the secondary screening task, and performs AES encryption on the known protein functional domains.

[0108] Understandably, a two-layer encryption mechanism provides a deeper level of data protection. In a specific implementation, the first layer of encryption targets task description information, including algorithm parameters, execution instructions, and return requirements, using a first temporary session key for encryption. The second layer of encryption targets core protein functional domain structural data, such as using AES encryption to encrypt sensitive data like the atomic coordinates and secondary structure information of known protein functional domains. This layered encryption strategy ensures both the confidentiality of task instructions and the security of core data.

[0109] Step S403: Distribute the encrypted task information and known protein functional domains to each slave node.

[0110] The distribution process is conducted through a trusted channel within the cluster, ensuring the security and reliability of the data transmission path. Optionally, an integrity verification mechanism can be added to the distribution process, where the master node generates a digital signature for each subtask, and the slave nodes verify the validity of the signature upon receipt, ensuring that the task data has not been tampered with during transmission.

[0111] Through the above-mentioned encrypted transmission mechanism, this embodiment achieves security protection throughout the entire task distribution process, which not only prevents the risk of leakage of sensitive data during the transmission of data within the cluster, but also ensures the integrity and authenticity of task instructions.

[0112] In one exemplary embodiment, such as Figure 5 As shown, in step S204, the protein data to be recompiled is correlated and compared with the target replacement protein data, and the boundary of the target functional domain of the target replacement protein data is adjusted. Specifically, this includes the following steps S501 to S503:

[0113] In step S501, the master node of the TEE server cluster splits the correlation comparison task into multiple comparison sub-tasks and distributes them to each slave node.

[0114] In practical applications, protein structure alignment computation is highly complex, and distributed computing can significantly improve processing efficiency. Therefore, within a TEE secure environment, the master node divides the overall alignment task into multiple independent alignment subtasks based on the number and structural complexity of the target replacement proteins. In one embodiment, each subtask contains a set of alignment datasets of the target replacement proteins and the proteins to be recompiled, along with a preset parameter template for the Dali algorithm, including key parameters such as the distance matrix construction threshold and alignment scoring matrix. Before task allocation, the master node also employs a secure encryption mechanism to encrypt both the task data and the protein data.

[0115] In step S502, each slave node calls the Dali distance matrix comparison algorithm to construct a distance matrix and align the structure of the target replacement protein and the protein to be reprogrammed.

[0116] After receiving encrypted subtask data within the TEE secure environment, the node decrypts the task data and loads the pre-compiled Dali algorithm module. Specifically, it first constructs a distance matrix between the functional domains of the known protein and the target replacement protein, calculating the spatial Euclidean distance between each residue pair; then, a dynamic programming algorithm is used to find the optimal structural alignment path and determine the correspondence between residues. The output of the Dali algorithm includes key indicators such as the residue correspondence of structural alignment, structural similarity score, residue-level distance matrix, and root mean square deviation.

[0117] Step S503: Based on the results of the Dali distance matrix alignment algorithm, analyze the spatial position matching degree of functional domain boundary residues, adjust the functional domain boundary interval of the target replacement protein, and obtain the functional domain boundary adjustment result set.

[0118] It is important to note that the initial boundary regions of the candidate functional domains selected by FoldSeek and TM-align are predefined based on the structural and sequence context of the domain in its original protein. When the Dali algorithm performs fine-grained 3D structural alignment between the target replacement protein containing the candidate functional domain and the protein to be reprogrammed containing the known functional domain, the goal is to find the residue correspondence that best matches the two in 3D space. This optimal alignment path may not terminate exactly at the initially defined boundary of the candidate functional domain, resulting in boundary extension, boundary contraction, or boundary shift.

[0119] Therefore, the analysis results output by the Dali algorithm may not be entirely consistent with the candidate functional domain intervals initially extracted from the TED database and defined based on the context of the source protein. Forcibly replacing these intervals with the original ones may lead to structural conflicts or functional abnormalities in the replaced proteins. Therefore, the adjustment of functional domain intervals also needs to be based on a functional comparison between known functional domains and candidate functional domains.

[0120] In one specific implementation, the node analyzes the spatial matching degree of residues in the boundary region based on the residue correspondence and spatial matching information provided by the Dali algorithm, and performs necessary expansion or contraction adjustments to the functional domain boundaries. For example, the adjustment range is typically controlled within 1-3 residues to ensure that the replaced functional domain can smoothly connect with the original protein structurally, and to avoid the risk of decreased protein structural stability or loss of function due to inaccurate boundary definition. The functional domain boundary adjustment results based on the Dali algorithm are shown below. Figure 6 As shown.

[0121] In some other embodiments, to ensure the accuracy of the boundary adjustment results, the boundary adjustment process can also be combined with a multi-node cross-validation mechanism. Specifically, after the slave node completes the functional domain boundary adjustment and returns the encrypted result, the master node performs the following verification process within the TEE secure environment:

[0122] First, the master node decrypts and aggregates the encrypted boundary adjustment results returned by each slave node, forming a preliminary functional domain boundary adjustment result set. The aggregation process includes data parsing, format standardization, and result classification to ensure the comparability of data returned by different nodes.

[0123] When the same alignment task is assigned to multiple slave nodes, the master node compares and analyzes the boundary adjustment results returned by different nodes, calculating the consistency ratio among the results. Specifically, the master node parses the boundary adjustment results of the same target replacement protein in different slave nodes, comparing their functional domain start / end residue positions. If multiple nodes agree on the functional domain boundaries of the same protein (e.g., the difference in residue positions is within a preset tolerance range), the result is deemed to have passed verification. If there are differences between the results, the system determines that the verification has failed, marks it as data to be checked, and filters or records it in the system log for subsequent manual review.

[0124] Optionally, the preset threshold for consistency verification can be configured according to actual needs. Alternatively, a confidence assessment mechanism can be introduced during the verification process to automatically calculate the confidence score for each boundary adjustment result based on the consistency level of the results, providing a more refined reference for subsequent functional domain replacements. When the same comparison task is assigned to multiple slave nodes, the system requires the consistency of the output results from each node to reach a preset threshold, thereby further improving the reliability of the boundary adjustment results. All functional domain boundary adjustment results are encrypted locally on the slave nodes and then returned to the master node through a trusted channel within the cluster.

[0125] Through the aforementioned distributed comparison and boundary adjustment mechanism, this embodiment effectively improves the accuracy and reliability of functional domain boundary adjustment results while ensuring data security, and reduces the deviation that may occur in single-node calculations. This provides a foundation for subsequent functional domain replacement operations and effectively solves the structural conflict problem caused by inaccurate boundary positioning in traditional methods.

[0126] To ensure the security of data interaction during the adjustment process, this embodiment employs an encrypted transmission mechanism in the task distribution stage. Specifically, as follows: Figure 7 As shown, in step S501, the process by which the master node splits the association comparison task into multiple comparison sub-tasks and distributes them to each slave node is achieved through the following encryption steps S601 to S603:

[0127] Step S601: The master node generates a second temporary session key for each slave node and transmits the second temporary session key to the slave node through the cluster internal key exchange mechanism.

[0128] The master node first generates an independent second temporary session key for each slave node. It should be noted that this second temporary session key is isolated from the first temporary session key used for similarity retrieval in step S401 to prevent the risk of key reuse. The master node securely transmits the second temporary session key to the corresponding slave node through the cluster's internal key exchange mechanism. In one specific implementation, this mechanism employs asymmetric encryption; the master node uses the public key of each slave node to encrypt the second temporary session key, ensuring that only the target slave node can decrypt and obtain it using its private key.

[0129] Step S602: A layered encryption mechanism is adopted, comparing the task information of the sub-tasks with public key encryption, and encrypting the target replacement protein and the protein to be reprogrammed with a second temporary session key.

[0130] Furthermore, the system employs a layered encryption strategy for task metadata and core data. For task description information of the alignment subtasks, including metadata such as the number of alignment groups, execution priority, and algorithm parameters, encryption is performed using the master node's public key; while for the core structural data of the target replacement protein and the protein to be reprogrammed, such as the Cα atom coordinate matrix and residue sequences, encryption is performed using a second temporary session key.

[0131] Step S603: Distribute the encrypted task information, target replacement protein, and protein to be reprogrammed to each slave node.

[0132] Through the above-mentioned encrypted transmission mechanism, this embodiment achieves security protection for the entire process of comparison task distribution. Through the layered encryption mechanism, it not only ensures the secure transmission of core sensitive data, but also ensures that each slave node can only obtain the sub-task data it is responsible for, thus achieving secure isolation between tasks.

[0133] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0134] Based on the same inventive concept, this application also provides a protein functional domain reprogramming apparatus based on a TEE server cluster for implementing the methods described above. The solution provided by this apparatus is similar to the implementation scheme described in the above methods; therefore, specific limitations in one or more apparatus embodiments provided below can be found in the limitations of the protein functional domain method based on a TEE server cluster described above, and will not be repeated here.

[0135] In one exemplary embodiment, such as Figure 8 As shown, a protein functional domain reprogramming device 700 based on a TEE server cluster is provided. The main operations of the protein functional domain method are performed by the TEE server cluster 104. The device includes: a data receiving module 701, a similar domain retrieval module 702, a protein acquisition module 703, a boundary adjustment module 704, a domain replacement processing module 705, and a result return module 706, wherein:

[0136] Data receiving module 701 is used to receive protein data to be reprogrammed sent by a client, wherein the protein data to be reprogrammed contains information about known functional domains;

[0137] The similarity domain retrieval module 702 is used to perform similarity retrieval on the known functional domains of the protein data to be recompiled based on the TED database, and to obtain target functional domains that are similar to the known functional domains.

[0138] The protein acquisition module 703 is used to acquire target substitution protein data corresponding to the target functional domain from the protein sequence database and the protein structure database.

[0139] The boundary adjustment module 704 is used to correlate and compare the protein data to be recompiled with the target replacement protein data, adjust the target functional domain boundary of the target replacement protein data, and obtain the target functional domain boundary adjustment result.

[0140] The domain replacement processing module 705 is used to replace the known functional domains in the protein data to be recompiled according to the target functional domain boundary adjustment result, and generate functional domain recompiled protein data.

[0141] The result return module 706 is used to return the protein data after the functional domain reprogramming to the client.

[0142] In an exemplary embodiment, the above-described apparatus includes, in addition to the TEE server cluster, a client 102 and a third-party verification server 106, which work together to complete a secure functional domain reprogramming process.

[0143] Client 102 is used to encrypt the protein data to be recompiled, initiate functional domain recompilation operation requests to the TEE server cluster, and obtain the functional domain recompiled protein data.

[0144] The third-party verification server 106 is used to receive and verify the overall trustworthiness certificate of the TEE server cluster sent by the TEE server cluster to confirm the trustworthiness of the TEE server cluster.

[0145] Specifically, client 102 first encrypts the protein data to be recompiled and initiates a functional domain recompilation request to the TEE server cluster. The master node of the TEE server cluster verifies the integrity of the TEE environment of each node through an internal trusted channel, generates a cluster-wide trusted certificate, and sends it to a third-party verification server 106 for verification.

[0146] After completing the verification, the third-party verification server 106 returns a trusted confirmation of the TEE cluster environment to the client 102. The client 102 then transmits the encrypted data to the TEE server cluster through a secure transmission channel. After the TEE server cluster completes the recompiling calculation of all functional domains within the secure environment, the client 102 obtains the result through the authorization mechanism.

[0147] Specifically, obtaining results through an authorization mechanism, such as Figure 9 As shown, client 102 submits its identity and task credentials to the master node. After verification by the TEE remote authentication mechanism and the third-party verification server 106, it obtains a one-time authorization credential. Client 102 uses this credential to initiate a result retrieval request. After verifying the validity of the credential, the master node transmits the encrypted result to client 102 through a secure transmission channel.

[0148] Through the aforementioned multi-party security collaboration mechanism, this embodiment achieves end-to-end trust assurance from task initiation to result return, ensuring the technical effectiveness of data security and computational reliability during protein functional domain reprogramming.

[0149] Each module in the aforementioned protein functional domain reprogramming device based on a TEE server cluster can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0150] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0151] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0152] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A protein functional domain reprogramming method based on a TEE server cluster, characterized in that, The TEE server cluster stores a TED database, a protein sequence database, and a protein structure database. The method is executed by the TEE server cluster and includes: Receive protein data to be reprogrammed sent by the client, wherein the protein data to be reprogrammed contains information about known functional domains; Based on the TED database, a similarity search is performed on the known functional domains of the protein data to be recompiled to obtain target functional domains that are similar to the known functional domains. Obtain target substitution protein data corresponding to the target functional domain from the protein sequence database and the protein structure database; The protein data to be reprogrammed is correlated and compared with the target replacement protein data, and the boundary of the target functional domain of the target replacement protein data is adjusted to obtain the target functional domain boundary adjustment result. Based on the target functional domain boundary adjustment results, the known functional domains in the protein data to be reprogrammed are replaced to generate functional domain reprogrammed protein data. The protein data after the functional domains have been reprogrammed is returned to the client.

2. The method according to claim 1, characterized in that, The step of performing a similarity search on known functional domains of the protein data to be recompiled based on the TED database to obtain target functional domains similar to the known functional domains includes: The master node of the TEE server cluster splits the similarity retrieval task into a preliminary screening sub-task based on the FoldSeek algorithm and a fine screening sub-task based on the TM-align algorithm, and distributes them to different slave nodes. The slave nodes that perform the initial screening subtask use the FoldSeek algorithm to perform the initial comparison and obtain the initial screening results of similar functional domains; The slave node executing the fine screening task then uses the TM-align algorithm to perform a second comparison of the initial screening results, and selects the target functional domains whose similarity to the known functional domains reaches a preset threshold. The master node summarizes the retrieval results of the slave nodes that performed the fine screening task.

3. The method according to claim 2, characterized in that, After the master node summarizes the retrieval results of the slave nodes that performed the fine screening sub-task, the method further includes: The master node uses a dual-algorithm cross-validation mechanism to filter out functional domains that are only recognized by a single algorithm and whose similarity is below a preset threshold, thus forming verified similar functional domains.

4. The method according to claim 2, characterized in that, The master node splits the similar functional domain retrieval task into a preliminary screening sub-task based on the FoldSeek algorithm and a fine screening sub-task based on the TM-align algorithm, and distributes them to different slave nodes, including: The master node generates a first temporary session key for each of the slave nodes and transmits the first temporary session key to the slave nodes through asymmetric encryption. A two-layer encryption mechanism is adopted, which encrypts the task information of the initial screening sub-task and the fine screening sub-task using the first temporary session key, and performs AES encryption on the known functional domain; The encrypted task information and the known functional domains are distributed to each slave node.

5. The method according to claim 1, characterized in that, The step of associating and comparing the protein data to be recompiled with the target replacement protein data, and adjusting the target functional domain boundary of the target replacement protein data to obtain the target functional domain boundary adjustment result includes: The master node of the TEE server cluster splits the association comparison task into multiple comparison sub-tasks and distributes them to each slave node. Each slave node calls the Dali distance matrix comparison algorithm to construct a distance matrix and align the structure of the target replacement protein and the protein to be reprogrammed. Based on the results of the Dali distance matrix alignment algorithm, the spatial position matching degree of functional domain boundary residues is analyzed, the functional domain boundary intervals of the target replacement protein are adjusted, and a functional domain boundary adjustment result set is obtained.

6. The method according to claim 5, characterized in that, After adjusting the functional domain boundary regions of the target replacement protein to obtain a functional domain boundary adjustment result set, the method further includes: The master node summarizes the functional domain boundary adjustment result set of each slave node; The master node adopts a multi-node cross-validation strategy to perform consistency verification on the functional domain boundary adjustment results obtained by distributing the same comparison task to different slave nodes, and obtains the functional domain boundary adjustment results whose consistency reaches a preset threshold.

7. The method according to claim 5, characterized in that, The master node of the TEE server cluster splits the comparison task into multiple comparison sub-tasks and distributes them to each slave node, including: The master node generates a second temporary session key for each of the slave nodes and transmits the second temporary session key to the slave nodes through the cluster internal key exchange mechanism; A layered encryption mechanism is adopted, in which the task information of the alignment subtask is encrypted with a public key, and the target replacement protein and the protein to be reprogrammed are encrypted with the second temporary session key. The encrypted task information, the target replacement protein, and the protein to be reprogrammed are distributed to each slave node.

8. The method according to claim 1, characterized in that, The TED database, the protein sequence database, and the protein structure database are stored in the TEE server cluster using a distributed encrypted storage strategy, which includes: The master node of the TEE server cluster uses the Gocryptfs tool to encrypt the TED database, the protein sequence database, and the protein structure database, and performs fragmentation processing to obtain encrypted fragmented data. The master node stores the encrypted shard data to the slave node and saves an encrypted storage mapping table in the master node's TEE environment. The storage mapping table records the correspondence between the encrypted shard data and the slave node.

9. A protein functional domain reprogramming device based on a TEE server cluster, characterized in that, The TEE server cluster stores a TED database, a protein sequence database, and a protein structure database. The device is executed by the TEE server cluster and includes: The data receiving module is used to receive protein data to be reprogrammed sent by the client, wherein the protein data to be reprogrammed contains information about known functional domains; The similarity domain retrieval module is used to perform similarity retrieval on the known functional domains of the protein data to be recompiled based on the TED database, and to obtain target functional domains that are similar to the known functional domains. The protein acquisition module is used to acquire target substitution protein data corresponding to the target functional domain from the protein sequence database and the protein structure database. The boundary adjustment module is used to correlate and compare the protein data to be recompiled with the target replacement protein data, adjust the target functional domain boundary of the target replacement protein data, and obtain the target functional domain boundary adjustment result. The domain replacement processing module is used to replace the known functional domains in the protein data to be recompiled according to the target functional domain boundary adjustment results, and generate functional domain recompiled protein data. The result return module is used to return the protein data after the functional domain reprogramming to the client.

10. The apparatus according to claim 9, characterized in that, The device also includes the client and the third-party verification server: The client is used to encrypt the protein data to be recompiled, initiate a functional domain recompilation operation request to the TEE server cluster, and obtain the protein data after functional domain recompilation. The third-party verification server is used to receive and verify the overall trustworthiness certificate of the TEE server cluster sent by the TEE server cluster to confirm the trustworthiness of the TEE server cluster.

Images